Last week, our against the spread NFL picks went 7-6, and our two totals plays split. On the season, our published plays are 59-42-3 in aggregate, for a 56.7% win rate (spread picks are 41-30-2 and totals plays are 18-12-1).
This week's Thursday Night Football matchup featured the Detroit Lions at the Green Bay Packers. Aaron Rodgers delivered an exhilarating finish at the last play of the game, evading various Lions pass rushers well behind the line of scrimmage to ultimately fashion out a marvelous 61-yard Hail Mary pass to tight end Richard Rogers. That play cemented the win for the Packers (and a cover for us). Green Bay's 11th hour, come-from behind victory is doubly remarkable as the win-clinching play was only made possible by a controversial face-mask penalty against Detroit that gave rise to that final, game-changing snap.
Brian Burke, of Advanced Football Analytics, documented sometime ago the virtual futility of the Hail Mary pass from very deep. Burke found that, between 2000 and 2011, almost none of the attempts were completed from past the mid-field marker in situations where a touchdown was required to tie or win.
Given the improbably positive start to our NFL betting week, we choose to believe that the football Gods are smiling on us currently (we hope that the hours of data collection and numbers crunching we have committed to this endeavor do not hurt either). As such, it is with great excitement that we unveil our highest conviction plays for week 13 NFL Sunday:
We show in the graphic above our actual bets, as well as our system's estimate of the line and the forecast probability of our wager covering (based on a logistic regression model that incorporates all output from the individual power ranking and simulation inputs in our model, plus broader measures meant to gauge sentiment).
We note biases in favor of higher scores mainly, and secondarily, toward road teams and underdogs this week. We remarked last week that the shift in our model's preference from UNDERs to OVERs of late might be indicative of the market adapting to bettors' biases. For example, DiFilippo, Krieger, et al (2012) observed evidence that the unders are more likely to cash in early in the season than overs (this adge dissipates with timehowever). The authors found that the naive strategy of blindly siding with the under in the first week of the NFL season to be profitable over the period under examination. Granted we are now closer to the last week of the season than to the first, however, given the prominence of UNDER selections in our totals work (with reasonable success) until recently, and then the hard about-face over the last couple of weeks, the hypothesis appears theoretically viable (for now, anyway) that an extended predisposition to bet on higher scores might finally be waning.
Separately, away teams were 76-66 versus the number through week 10, demonstrating a nominally profitable edge. However, traveling squads went 13-15 over the last two weeks. A reversal of this stint of sub-par performance might likewise be a reasonable expectation from the current position.
And finally, the irrational tendency among bettors to back the favorite, even when the spread supports the underdog, is well documented in research literature. As such, our tilt this week in support of the underdog appears consistent with our objective of emotionless handicapping.
Happy betting!!